Paper Abstract and Keywords |
Presentation |
2022-07-15 11:30
Development and Evaluation of Automated Design System of Automotive Ethernet by Linkage of Machine Learning and Meta-heuristics Yasuhiro Mori, Hiroshi Yamamoto (Ritsumeikan Univ.), Yojiro Suyama, Tatsuya Izumi (Sumitomo Electric Industries), Hirofumi Urayama (AutoNetwork Technology Research Institute), Shiho Kobayashi, Shigeki Umehara, Hideaki Tani (Sumitomo Electric System Solutions) NS2022-54 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
In recent years, Ethernet tends to be adopted as an in-vehicle network that the various computers (ECUs) are interconnected. However, there is concern that the packets concentrate on certain switches, which become bottlenecks causing packet losses and lack of real-time property of the important data. Therefore, in this study, we propose a method to automatically derive appropriate QoS control settings to avoid the occurrence of the bottlenecks based on the network configuration and the characteristics of the network traffic. For the proposed method, a machine learning model is built to estimate the presence and degree of bottlenecks by learning the results of simulations carried out in advance on various network topologies, types of traffic and QoS control settings. Furthermore, by utilizing a meta-heuristic technique, an automated design system of automotive Ethernet is realized to automatically derive the appropriate QoS control settings for avoiding the bottleneck. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
IoT / In-vehicle Ethernet / QoS controll / Machine Learning / Meta-heuristic / / / |
Reference Info. |
IEICE Tech. Rep., vol. 122, no. 105, NS2022-54, pp. 131-136, July 2022. |
Paper # |
NS2022-54 |
Date of Issue |
2022-07-06 (NS) |
ISSN |
Online edition: ISSN 2432-6380 |
Copyright and reproduction |
All rights are reserved and no part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Notwithstanding, instructors are permitted to photocopy isolated articles for noncommercial classroom use without fee. (License No.: 10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) |
Download PDF |
NS2022-54 |